Fix VDMS retrieval issue (opea-project#1252)

* Fix VDMS retrieval issue
Signed-off-by: lvliang-intel <[email protected]>

comps/retrievers/src/Dockerfile

@@ -26,7 +26,7 @@ RUN pip install --no-cache-dir --upgrade pip setuptools && \
     fi && \
     pip install --no-cache-dir torch torchvision ${PIP_EXTRA_INDEX_URL} && \
     pip install --no-cache-dir ${PIP_EXTRA_INDEX_URL} -r /home/user/comps/retrievers/src/requirements.txt && \
-    pip install opentelemetry-api==1.27.0 opentelemetry-exporter-otlp==1.27.0 opentelemetry-sdk==1.27.0
+    pip install opentelemetry-api==1.29.0 opentelemetry-exporter-otlp==1.29.0 opentelemetry-sdk==1.29.0
 
 ENV PYTHONPATH=$PYTHONPATH:/home/user
 

comps/retrievers/src/integrations/vdms.py

@@ -48,7 +48,7 @@ def _initialize_embedder(self):
             from comps.third_parties.clip.src.clip_embedding import vCLIP
 
             embeddings = vCLIP({"model_name": "openai/clip-vit-base-patch32", "num_frm": 64})
-        if TEI_EMBEDDING_ENDPOINT:
+        elif TEI_EMBEDDING_ENDPOINT:
             # create embeddings using TEI endpoint service
             if logflag:
                 logger.info(f"[ init embedder ] TEI_EMBEDDING_ENDPOINT:{TEI_EMBEDDING_ENDPOINT}")

comps/retrievers/src/requirements.txt

@@ -3,6 +3,7 @@ cairosvg
 docarray[full]
 docx2txt
 easyocr
+einops
 fastapi
 future
 graspologic

comps/third_parties/pathway/deployment/docker_compose/compose.yaml

@@ -2,7 +2,7 @@
 # SPDX-License-Identifier: Apache-2.0
 
 include:
-   - ../../../tei/deployment/docker_compose/compose.yaml
+  - ../../../tei/deployment/docker_compose/compose.yaml
 
 services:
   pathway-db:
@@ -12,13 +12,15 @@ services:
       - "${PATHWAY_PORT:-6379}:${PATHWAY_PORT:-6379}"
     volumes:
       - "${PATHWAY_VOLUME:-../../src/README.md}:/app/data/README.md"
-    network_mode: host
     environment:
+      no_proxy: ${no_proxy}
       http_proxy: ${http_proxy}
       https_proxy: ${https_proxy}
       PATHWAY_HOST: ${PATHWAY_HOST_DB}
       PATHWAY_PORT: ${PATHWAY_PORT}
       TEI_EMBEDDING_ENDPOINT: ${TEI_EMBEDDING_ENDPOINT}
+      HF_TOKEN: ${HF_TOKEN}
+      HUGGINGFACEHUB_API_TOKEN: ${HF_TOKEN}
     healthcheck:
       test: ["CMD-SHELL", "sleep 30 && exit 0"]
       interval: 1s

comps/third_parties/pathway/src/requirements.txt

@@ -1,7 +1,7 @@
 langchain
 langchain-community
-langchain_huggingface
-langchain_openai
+openai
 pathway[xpack-llm]
 sentence-transformers
+tiktoken
 unstructured[all-docs] >= 0.16

comps/third_parties/pathway/src/vectorstore_pathway.py

@@ -7,8 +7,7 @@
 import nltk
 import pathway as pw
 from langchain import text_splitter
-from langchain_community.embeddings import HuggingFaceBgeEmbeddings
-from langchain_huggingface import HuggingFaceEndpointEmbeddings
+from langchain_community.embeddings import HuggingFaceBgeEmbeddings, HuggingFaceInferenceAPIEmbeddings
 from pathway.xpacks.llm.parsers import ParseUnstructured
 from pathway.xpacks.llm.vector_store import VectorStoreServer
 
@@ -40,15 +39,17 @@
 port = int(os.getenv("PATHWAY_PORT", 8666))
 
 EMBED_MODEL = os.getenv("EMBED_MODEL", "BAAI/bge-base-en-v1.5")
-
+HUGGINGFACEHUB_API_TOKEN = os.getenv("HUGGINGFACEHUB_API_TOKEN", "")
 tei_embedding_endpoint = os.getenv("TEI_EMBEDDING_ENDPOINT")
 
 if __name__ == "__main__":
     # Create vectorstore
     if tei_embedding_endpoint:
         # create embeddings using TEI endpoint service
         logging.info(f"Initializing the embedder from tei_embedding_endpoint: {tei_embedding_endpoint}")
-        embeddings = HuggingFaceEndpointEmbeddings(model=tei_embedding_endpoint)
+        embeddings = HuggingFaceInferenceAPIEmbeddings(
+            api_key=HUGGINGFACEHUB_API_TOKEN, model_name=EMBED_MODEL, api_url=tei_embedding_endpoint
+        )
     else:
         # create embeddings using local embedding model
         embeddings = HuggingFaceBgeEmbeddings(model_name=EMBED_MODEL)

tests/retrievers/test_retrievers_elasticsearch.sh

@@ -79,6 +79,8 @@ function validate_microservice() {
 function stop_docker() {
     cd $WORKPATH/comps/retrievers/deployment/docker_compose
     docker compose -f compose.yaml down  ${service_name} --remove-orphans
+    cid=$(docker ps -aq --filter "name=elasticsearch-vector-db")
+    if [[ ! -z "$cid" ]]; then docker stop $cid && docker rm $cid && sleep 1s; fi
 }
 
 function main() {

tests/retrievers/test_retrievers_milvus.sh

@@ -83,6 +83,8 @@ function stop_docker() {
 
     cd $WORKPATH/comps/retrievers/deployment/docker_compose
     docker compose -f compose.yaml down  ${service_name} --remove-orphans
+    cid=$(docker ps -aq --filter "name=tei-embedding-serving")
+    if [[ ! -z "$cid" ]]; then docker stop $cid && docker rm $cid && sleep 1s; fi
 }
 
 function main() {

tests/retrievers/test_retrievers_neo4j_on_intel_hpu.sh

@@ -46,8 +46,8 @@ function start_service() {
     export RETRIEVER_PORT=11635
     export EMBEDDING_MODEL_ID="BAAI/bge-base-en-v1.5"
     export DATA_PATH="/data2/cache"
-    export MAX_INPUT_TOKENS=1024
-    export MAX_TOTAL_TOKENS=3000
+    export MAX_INPUT_TOKENS=4096
+    export MAX_TOTAL_TOKENS=8192
     export TEI_EMBEDDING_ENDPOINT="http://${host_ip}:${TEI_EMBEDDER_PORT}"
     export LLM_MODEL_ID="meta-llama/Meta-Llama-3.1-8B-Instruct"
     export TGI_LLM_ENDPOINT="http://${host_ip}:${LLM_ENDPOINT_PORT}"
@@ -67,7 +67,7 @@ function start_service() {
     docker run -d --name="test-comps-retrievers-neo4j-llama-index-dataprep" -p 6004:5000 -v ./data:/data --ipc=host -e TGI_LLM_ENDPOINT=$TGI_LLM_ENDPOINT \
         -e TEI_EMBEDDING_ENDPOINT=$TEI_EMBEDDING_ENDPOINT -e TEI_ENDPOINT=$TEI_EMBEDDING_ENDPOINT -e EMBEDDING_MODEL_ID=$EMBEDDING_MODEL_ID -e LLM_MODEL_ID=$LLM_MODEL_ID -e host_ip=$host_ip -e no_proxy=$no_proxy \
         -e http_proxy=$http_proxy -e https_proxy=$https_proxy -e NEO4J_URL="bolt://${host_ip}:${NEO4J_PORT2}" -e NEO4J_USERNAME="neo4j" \
-        -e NEO4J_PASSWORD="neo4jtest" -e HF_TOKEN=$HF_TOKEN -e LOGFLAG=True -e DATAPREP_COMPONENT_NAME="OPEA_DATAPREP_NEO4J_LLAMAINDEX" opea/dataprep-neo4j-llamaindex:comps
+        -e NEO4J_PASSWORD="neo4jtest" -e HF_TOKEN=$HF_TOKEN -e MAX_INPUT_LEN=$MAX_INPUT_TOKENS -e LOGFLAG=True -e DATAPREP_COMPONENT_NAME="OPEA_DATAPREP_NEO4J_LLAMAINDEX" opea/dataprep-neo4j-llamaindex:comps
 
     sleep 1m
 
@@ -152,7 +152,7 @@ function validate_microservice() {
 }
 
 function stop_docker() {
-    cid=$(docker ps -aq --filter "name=test-comps-*")
+    cid=$(docker ps -aq --filter "name=test-comps-*" --filter "name=neo4j-apoc" --filter "name=tgi-gaudi-server" --filter "name=tei-embedding-serving")
     if [[ ! -z "$cid" ]]; then docker stop $cid && docker rm $cid && sleep 1s; fi
     cd $WORKPATH/comps/retrievers/deployment/docker_compose
     docker compose -f compose.yaml down  ${service_name} --remove-orphans

tests/retrievers/test_retrievers_opensearch.sh

@@ -75,6 +75,8 @@ function validate_microservice() {
 function stop_docker() {
     cd $WORKPATH/comps/retrievers/deployment/docker_compose
     docker compose -f compose.yaml down  ${service_name} --remove-orphans
+    cid=$(docker ps -aq --filter "name=opensearch-vector-db" --filter "name=tei-embedding-serving")
+    if [[ ! -z "$cid" ]]; then docker stop $cid && docker rm $cid && sleep 1s; fi
 }
 
 function main() {

tests/retrievers/test_retrievers_pathway.sh

@@ -69,6 +69,8 @@ function validate_microservice() {
 function stop_docker() {
     cd $WORKPATH/comps/retrievers/deployment/docker_compose
     docker compose -f compose.yaml down  ${service_name} --remove-orphans
+    cid=$(docker ps -aq --filter "name=pathway-db")
+    if [[ ! -z "$cid" ]]; then docker stop $cid && docker rm $cid && sleep 1s; fi
 }
 
 function main() {

tests/retrievers/test_retrievers_pgvector.sh

@@ -64,6 +64,8 @@ function validate_microservice() {
 function stop_docker() {
     cd $WORKPATH/comps/retrievers/deployment/docker_compose
     docker compose -f compose.yaml down  ${service_name} --remove-orphans
+    cid=$(docker ps -aq --filter "name=pgvector-db")
+    if [[ ! -z "$cid" ]]; then docker stop $cid && docker rm $cid && sleep 1s; fi
 }
 
 function main() {

tests/retrievers/test_retrievers_qdrant.sh

@@ -59,6 +59,8 @@ function validate_microservice() {
 function stop_docker() {
     cd $WORKPATH/comps/retrievers/deployment/docker_compose
     docker compose -f compose.yaml down  ${service_name} --remove-orphans
+    cid=$(docker ps -aq --filter "name=qdrant-vector-db")
+    if [[ ! -z "$cid" ]]; then docker stop $cid && docker rm $cid && sleep 1s; fi
 }
 
 function main() {

tests/retrievers/test_retrievers_redis.sh

@@ -131,6 +131,8 @@ function validate_mm_microservice() {
 function stop_docker() {
     cd $WORKPATH/comps/retrievers/deployment/docker_compose
     docker compose -f compose.yaml down  ${service_name} ${service_name_mm} --remove-orphans
+    cid=$(docker ps -aq --filter "name=redis-vector-db")
+    if [[ ! -z "$cid" ]]; then docker stop $cid && docker rm $cid && sleep 1s; fi
 }
 
 function main() {

tests/retrievers/test_retrievers_vdms.sh

@@ -78,6 +78,8 @@ function validate_microservice() {
 function stop_docker() {
     cd $WORKPATH/comps/retrievers/deployment/docker_compose
     docker compose -f compose.yaml down  ${service_name} ${service_name_mm} --remove-orphans
+    cid=$(docker ps -aq --filter "name=retriever-vdms*" --filter "name=vdms-vector-db" --filter "name=tei-embedding-serving")
+    if [[ ! -z "$cid" ]]; then docker stop $cid && docker rm $cid && sleep 1s; fi
 }
 
 function main() {